The present invention relates to using distributed redundant array of independent disks (RAID), and more specifically this invention relates to transferring data from a traditional RAID structure to a distributed RAID structure.
An amount of time for a traditional RAID array rebuild to complete will never be better than drive capacity and/or drive bandwidth available in the traditional RAID. Consequently, traditional RAID array rebuild times are increasing quickly, with 3 TB nearline drives having a rebuild time in the range from about 10 hours to about 20 hours or more. The mean time between failure (MTBF) rates for RAID are decreasing because RAID are exposed to data loss due to medium errors or a second drive failure during the rebuilding process. Many users, especially users with a large number of arrays now consider RAID-5 to provide inadequate protection against data loss.
Therefore, should a RAID user want to use distributed RAID in an existing system, then data migration must be performed which needs more time and brings more risk of failure or data loss to the whole system.